A version of this document hosted on GitHub Pages can be found here.
—Executive Summary—
National public health messengers have become politicized during the COVID-19 pandemic, creating a vaccuum for local public health messengers to communicate information about the virus and its spread. Surveys of public trust find that Americans trust their local governments more than they trust the federal government. Many local governments have social media accounts on various platforms, in order to reach wider audiences and more effectively reach their citizens. The visualizations presented here indicate that counties whose public health departments have Facebook or Twitter accounts have, on average, lower rates of COVID-19 cases and deaths and higher rates of COVID-19 vaccination than counties whose public health departments do not have Facebook or Twitter accounts.
In times of public health crisis, it is essential that officials and science communicators are able to reach as many people as possible to inform them of mitigation and prevention measures, research updates, available resources, and new laws regarding individual or organizational activity. The emergence and spread of SARS-CoV-2 (the virus which causes COVID-19) presents an interesting and novel phenomenon; it is the first global pandemic during the social media age. Prior epidemics this century, like SARS in 2002, H1N1 flu in 2009, and Ebola in 2014, either happened when social media platforms were young and not widely used, or were geographically isolated or limited in spread by pathogen-specific factors. Public health communication in the United States faces an additional hurdle. The leader of one of the two major national political parties (and the immediate former president) actively and vocally opposed many of the suggestions offered by top-level health advisors, leading many in the country to become distrustful of government messaging around the virus. The polarization of discourse at the national level around COVID-19 means that local health officials could potentially function as more effective vectors of information. This is because a local health official could be a valued member of one’s community, while national figures are inherently several degrees removed from the day-to-day actions and considerations of the average citizen. Recent surveys have found a 20-point gap between self-reported trust of local government and trust in the federal government (Kettl, 2020). Other studies of the efficacy of social media messaging around public health find that recipients trust messaging from organizations more than messaging from individuals (Freberg, 2012). Local public health departments exist at the confluence of these two factors, and so are uniquely positioned to potentially influence public health outcomes.
The aim of this analysis is to investigate if counties whose public health departments have social media accounts on one or both of two major platforms (Facebook and Twitter) have different outcomes with respect to COVID-19 than counties whose health departments do not have social media accounts. After manually entering the data for this data set, I coded each county (or county-equivalent jurisdiction)1 as either having a Facebook account for its public health department, a Twitter account, both, or neither. I then located and incorporated data on cases, deaths, completed vaccinations, and first doses administered. Those values were then divided by county population to produce proportions. Figure 1 shows the average rates of each metric, broken down by county health department social media presence.
Of the 3,143 counties in the United States, 800 have both Facebook and Twitter accounts for their health departments. 1,375 have Facebook pages but not Twitter accounts, 136 have Twitter accounts but not Facebook pages, and 908 have neither. These are the groups represented in Figure 1 below.
Clicking on one of the legend entries will hide the corresponding line on each subplot, and the slider at the bottom can be used to adjust the range of dates to any interval between January 21, 2020 and December 14, 2021.
While all four groups of counties appear similar on Figure 1, there are small but notable gaps between them. Counties with no health department social media presence consistently have average case and death rates which are higher than those with Facebook or Twitter accounts, and the gap is largest between counties with both Facebook and Twitter pages for their health departments and counties with neither.
However, some counties whose health departments lack social media presences are represented online in other ways. In many parts of the country, particularly in rural states such as Idaho, Nebraska, and Kentucky, groups of adjacent counties are organized into multi-county public health districts, and these multi-county districts often have social media accounts. For the purposes of this project, counties in these organized districts are coded as having health department social media accounts, and are included in the relevant groups. The map below shows which counties and districts have which social media accounts. As part of the data collection process, I used the twint module for Python to scrape tweets from health department twitter accounts. Some accounts were unable to be scraped or had no available tweets, but I was able to retrieve a large number of tweets. For many of the counties and districts which have twitter accounts, clicking on them will produce a pop-up which includes a link to that account’s most recent tweet.
Counties which are part of multi-county health districts have been consolidated into those districts. The icon in the top-left corner can be used to select and de-select different layers of the map. For the best viewing experience, select one layer at a time. To see which counties are contained within a multi-county district, select the “Health District Social Media” layer and click on the district.